Created attachment 104174 [details]
issue.pdf

Created attachment 104176 [details]
issue.png

Created attachment 104308 [details]
Draw import of complex pdf

Test System:
Kubuntu 14.04 (KDE 4.13.2)
LibreOffice 4.3.0.3

The issue of Draw importing the horizontal, hatched stripes as a gray diagonal block occurs on my test system.

On LibreOffice 4.2.4.2, a similar import issue for the same hatched areas displays the hatching in diagonal lines. png image attached.

Thanks for reporting.
--Algot

The "shaded" areas are the cross hatching layer as applied to elements of the source document for the PDF. They should be hidden below other elements--but are visible when rendered in Draw.

Unclear what program Micron prepared the document in and used for export to PDF. 

When the document is opened in Adobe Reader, Adobe Acrobat, Adobe Photoshop, as well as with Inkscape--the hatching layer is correctly confined--and is not available as an object (e.g. all layers have been flattened). Only Adobe Illustrator did not recognize the format of the PDF.

As is, only LibreOffice incorrectly is showing the hatching layer but it is incorrect--if only a corner case.

On Windows 7 sp1, 64-bit en-US with
Version: 4.3.0.4
Build ID: 62ad5818884a2fc2e5780dd45466868d41009ec0

(In reply to comment #4)
> Unclear what program Micron prepared the document in and used for export to
> PDF. 

AH Formatter V5.3 R1 (5,3,2011,0425) for Windows
http://www.antennahouse.com/product/ahf50/ahf5top.htm

An "original" PDF can be found here:
http://www.micron.com/-/media/Documents/Products/Data%20Sheet/DRAM/DDR3/1Gb_DDR3_SDRAM.pdf

The version that I have attached to this bug report is an extract of a few pages.

(In reply to comment #5)
> An "original" PDF can be found here:
> http://www.micron.com/-/media/Documents/Products/Data%20Sheet/DRAM/DDR3/
> 1Gb_DDR3_SDRAM.pdf

Extracted the pages shown in attachment 104174 [details] (issue.pdf) but using Adobe Acrobat 9.5.5

When opened in Draw with a current LibreOffice Version: 4.4.0.0.alpha0+
Build ID: 32ce5ae15a8f156b4681c36d248b6731df3457c6
TinderBox: Win-x86@42, Branch:master, Time: 2014-08-25_01:27:12

The same unmasked view of the hatching layers are incorrectly being exposed in the PDF rendering in Draw.

This lines must be clipped. See Bug 86211

Bug still present in:

Version: 5.1.2.2
Build ID: 1:5.1.2~rc2-0ubuntu1~trusty0
CPU Threads: 8; OS Version: Linux 3.16; UI Render: default; 
Locale: de-DE (de_DE.UTF-8)

Bug still present in
Version: 6.0.0.0.alpha0+
Build-ID: 115bed941d7b7ed1b95d6424bfb98456c1d87546

Still reproducible with:
Version: 6.1.2.1
Build-ID: 1:6.1.2~rc1-0ubuntu0.14.04.4
CPU-Threads: 8; BS: Linux 4.4; UI-Render: Standard; VCL: gtk2; 
Gebietsschema: de-DE (de_DE.UTF-8); Calc: group threaded

Issue remains with the pdf import filters, still depends on bug 86211 for correct clipping mask

The pdfium based insert as image handle the masking correctly

Version: 7.2.2.1 (x64) / LibreOffice Community
Build ID: 0e408af0b27894d652a87aa5f21fe17bf058124c
CPU threads: 8; OS: Windows 10.0 Build 19043; UI render: Skia/Vulkan; VCL: win
Locale: en-US (en_US); UI: en-US
Calc: threaded

Created attachment 197717 [details]
More minimal pdf

I've been minimising this pdf by hand with the help of qpdf, and have got to this now tiny version.  It should have one lozenge and is still not clipped right.
This isn't as simple as a line clipping

The clip path is getting confused, because at one point it (correctly) should be empty; but the pdf import code is ignoring empty clip paths, so when the next clip path comes along, it uses this rather than staying empty.

So:

-    if( aCurClip.count() )  // #i92985# adapted API from (..., false, false) to (..., true, false)
-        aNewClip = basegfx::utils::clipPolyPolygonOnPolyPolygon( aCurClip, aNewClip, true, false );
+    aNewClip = basegfx::utils::clipPolyPolygonOnPolyPolygon( aCurClip, aNewClip, true, false );

helps things; but now we seem to have a new problem with some of the shading not being shown; so I need to track that down.

(A fun side part of this particular test file is it almost falls down my recent tilingPatternFill path which would have caused a blurred rendering, but that's a separate thing - I hadn't expected that to happen for what's closer to a shading)

Created attachment 197966 [details]
minimal file b

This is a different chunk of the pdf which exhibits a slightly different problem.

Making it not-ignore empty clip paths, and also closing incoming clip paths makes the 'More minimal pdf' from 22nd render OK, but it's not the whole story.
Looking at the whole doc, or in particular the 'minimal file b' I just uploaded; it's missing bits of the shading.

I think this is a winding rule screwup of the type tdf#86211 suggests.

Summarizing what we have in that minimal file b; we have a lozenge shape (which gives us a clip path to start with), and then we have square diamonds, each with it's own clip path, and inside those are a series of (11?) stroke lines.
Some of the diamonds end up losing all their stroke lines.

Here's some of my debug corresponding to the clip setup going into one of the (missing) square diamonds:


1439 PDFIProcessor::intersectClip: new:[1:<5:(-2.1378,-4.35001)--(18.8622,-4.35001)--(18.8622,16.65)--(-2.1378,16.65)--(-2.1378,-4.35001)>]
One of the clipping paths for a square as we read it
1440 PDFIProcessor::intersectClip: new(post close):[1:<5:(-2.1378,-4.35001)--(18.8622,-4.35001)--(18.8622,16.65)--(-2.1378,16.65)--(-2.1378,-4.35001)>]
1441 PDFIProcessor::intersectClip: new(post NZC):[1:<5:(-2.1378,-4.35001)--(18.8622,-4.35001)--(18.8622,16.65)--(-2.1378,16.65)--(-2.1378,-4.35001)>]
1442 PDFIProcessor::intersectClip:   new: xfrm: [1:<5:(38732.2,25528)--(40893.9,26878.9)--(39543.1,29040.6)--(37381.4,27689.8)--(38732.2,25528)>]
after going through a xfrm
1443 PDFIProcessor::intersectClip:   cur: [1:<6:(8486,28745.3)--(8486,27652.8)--(8490.6,27652.8)--(42524.4,27671)--(42838.9,28217.3)--(42524.4,28763.5)>]
'cur' is the clipping path before this call; this corresponds to the main lozenge
1444 PDFIProcessor::intersectClip:   res: [1:<5:(39717.2,28762)--(39096.7,28761.7)--(37381.4,27689.8)--(37394.8,27668.3)--(40399.7,27669.9)>]

OK, great - a valid clip path

1445 PDFIProcessor::intersectClip: new:[1:<4:(-2.1378,-4.35)--(18.862,-4.35)--(18.862,16.65)--(-2.1378,16.65)>]

We get called again, by an almost identical clip path - looks like poppler rounded it somewhere

1446 PDFIProcessor::intersectClip: new(post close):[1:<4:(-2.1378,-4.35)--(18.862,-4.35)--(18.862,16.65)--(-2.1378,16.65)>]

Note this wasn't closed, and so a call to basegfx::utils::closeWithGeometryChange  closes it (Hmm is it?)

1447 PDFIProcessor::intersectClip: new(post NZC):[1:<4:(-2.1378,-4.35)--(18.862,-4.35)--(18.862,16.65)--(-2.1378,16.65)>]
1448 PDFIProcessor::intersectClip:   new: xfrm: [1:<4:(38732.2,25528)--(40893.9,26878.9)--(39543.1,29040.6)--(37381.4,27689.8)>]
1449 PDFIProcessor::intersectClip:   cur: [1:<5:(39717.2,28762)--(39096.7,28761.7)--(37381.4,27689.8)--(37394.8,27668.3)--(40399.7,27669.9)>]

This is the clip we generated on 1444 above

1450 PDFIProcessor::intersectClip:   res: [1:<5:(39717.2,28762)--(39096.7,28761.7)--(37381.4,27689.8)--(37394.8,27668.3)--(40399.6,27669.9)>]

OK, valid clip comes out

1451 PDFIProcessor::intersectClip: new:[1:<4:(-2.1378,-4.35)--(-2.1378,16.65)--(18.862,16.65)--(18.862,-4.35)>]

Another call from poppler - this is the interesting one; the only difference from 1445 above is the point order - this is the opposite

1452 PDFIProcessor::intersectClip: new(post close):[1:<4:(-2.1378,-4.35)--(-2.1378,16.65)--(18.862,16.65)--(18.862,-4.35)>]
1453 PDFIProcessor::intersectClip: new(post NZC):[1:<4:(-2.1378,-4.35)--(-2.1378,16.65)--(18.862,16.65)--(18.862,-4.35)>]
1454 PDFIProcessor::intersectClip:   new: xfrm: [1:<4:(38732.2,25528)--(37381.4,27689.8)--(39543.1,29040.6)--(40893.9,26878.9)>]
1455 PDFIProcessor::intersectClip:   cur: [1:<5:(39717.2,28762)--(39096.7,28761.7)--(37381.4,27689.8)--(37394.8,27668.3)--(40399.6,27669.9)>]
1456 PDFIProcessor::intersectClip:   res: [0:]

Empty clip! But note this is clipping the same thing but to only the opposite direction; so if we agreed what that meant
I think this should survive.

(Reads the PDF spec, ...)
I'm still convinced that clip should be none-empty; but I'm not sure it's a winding rule issue.
Yes hte point order is opposite - but that's not winding rule when we're talking about two separate paths; each path has only one crossing, so it doesn't matter if you're using non-zero or even/odd rules, they both say fill.
I need to dig more into the clipping code.

I'm currently stuck on this; I posted to the list asking details about the clipping function, and would appreciate any suggestions:

https://lists.freedesktop.org/archives/libreoffice/2024-December/092759.html

@Armin, any cycles to compare notes with Dave G.?

Created attachment 199734 [details]
Enable tiling pattern fill get better result

I try use inkscape open the minimal files, it show the shape fill with pattern.
and I checked `PDFOutDev::tilingPatternFill` may fallback to slow method use clippath, After comment out these code and we can get better result with tilingPatternFill. remaining issue only the pattern not same with original file.

(In reply to zllangty from comment #24)
> Created attachment 199734 [details]
> Enable tiling pattern fill get better result
> 
> I try use inkscape open the minimal files, it show the shape fill with
> pattern.
> and I checked `PDFOutDev::tilingPatternFill` may fallback to slow method use
> clippath, After comment out these code and we can get better result with
> tilingPatternFill. remaining issue only the pattern not same with original
> file.

Yep, I'd spotted they were tilingPatternFill; but it's a fun problem:
  a) I think the fallback is only taken because of a rounding comparison; my debug notes have:
    1250      if (xStep != nWidth || yStep != nHeight)
    (gdb) p xStep
    $1 = 21
    (gdb) p nWidth
    $2 = 20.999999999999996
    (gdb) p yStep
    $3 = 21
    (gdb) p nHeight
    $4 = 21.000000000000004

  so we can probably fix that easily

  b) But it feels like we should fix the clipping problem anyway!